DQ PC170713¶

• Stable running for several days
  • Started Monday 15th July at 17:17
  • Finished Wednesday 17th July at 10:15
  • 172747 DAQ events (~ 60k cosmic spills)
  • No evidence of memory leaks
  • Ended because rogers accidentally shut his laptop (!)
  • Running against lp:~maus-mlcr/maus/control-room-3/, which is now merged with trunk
• MongoDB database size is now capped at 200 MB, about 10 minutes of running
  • Results in much better stability from MongoDB
• Histogram updates are now done every 60 spills
  • TOF reducer was getting hung up on writing images to disk

TODO:

• It is very clear that the current system is not satisfactory
  • Reconstruct -> Make histograms -> SerialiseAsJson(Write to disk -> Read from disk) -> Write to disk via OutputPyImage -> Pass to web browser
• Write to disk is bottleneck
• Much functionality is hard to implement
  • Pause/rewind button
  • GUI manipulation of histograms
• Propose using system analogous to DAQ online monitor
  • Present multiple TSockets one for each reducer
  • Each socket pushes a TCanvas and associated metadata
  • "Viewer" TApplication that reads the TSocket and presents to user
  • Reconstruct -> Make histograms -> SerialiseAsROOT -> Pass to TApplication
• Discuss
  • Do we ever need to push anything more sophisticated than a TCanvas?
  • e.g. do we want to have detector specific buttons/menus in the TApplication?